The goal of this research is to explore the neural systems underlying two critical aspects of speech and lexical processing: how listeners perceive a stable phonetic percept despite many sources of variability (the object constancy/invariance problem) and how they successfully select the appropriate word candidate from the many words that share sound shape properties (the resolution of phonological competition). One series of experiments will investigate invariance for phonetic categories across vowel environments and different speakers, and invariance for the phonetic feature of voicing across variations in place and manner of articulation and manner of articulation across variations in voice and place of articulation. An fMRI-adaptation paradigm will be used to identify brain regions in the phonetic processing stream that show invariant neural responses under different sources of variability. Invariant neural responses in temporal areas would suggest that phonetic invariance derives from invariant acoustic patterns. Invariant neural responses in frontal areas including the inferior frontal gyrus (IFG), premotor, and motor areas would suggest that phonetic invariance derives from shared motor/articulatory gestures. Invariant neural responses in parietal areas would suggest that phonetic constancy derives from higher order abstract phonological representations. And invariant neural responses in the IFG in the absence of activation of articulatory areas would suggest that phonetic constancy is determined by decision-related processes. A second series of experiments will investigate the brain regions recruited under conditions of phonological competition. It is hypothesized that the IFG serves a domain general role and hence will be recruited under conditions of phonological competition. However, it is hypothesized that the IFG is not responsive to competition inherent in the phonological properties of the stimulus but is recruited only in the resolution of competition, i.e. when competing representations are activated and a word needs to be selected from among these competing representations. In contrast, it is hypothesized that the SMG is recruited in accessing the lexical form of a word irrespective of selection demands with activation modulated by the degree of phonological competition. Finally, it is proposed that phonological competition recruits the same neural systems in both auditory word recognition and spoken word production. In auditory word recognition, an eyetracking paradigm will be used to investigate onset competition, lexical density, and phonologically mediated competition. In spoken word production, word reading and picture naming paradigms will be used to investigate the production of words with and without voiced competitors, the production of words from high versus low density neighborhoods, and the production of words presented in the same phonological competitor set versus those not in the same competitor set.